I have made recordings of a single source (musical instrument) with multiple microphones and in the interest of limiting the size of the final project, I would like to find a way to get the impulse response between a ref microphone and the others.

I have tried doing this in python by simply taking the FFT of both signals (ref mic A -> mic B for example), dividing them and then doing the inverse FFT of the result to get the impulse response, but this simply does not work correctly, even though i see the center peak suggesting an impulse response, the result I get is very noisy and highly inaccurate probably because that process tries to compare the delay and eq of the noise and it just creates peaks everywhere.

I stumbled upon these questions/answers: System identification/ Filter estimation to mimic frequency equalizer of audio with Scipy and Compensating Loudspeaker frequency response in an audio signal

And it looks like it's doing exactly what i need (retrieving both phase change and frequency response), my only problem being that i'm not very good at signal processing (yet) and i don't know what to do with that "equalizer" and more specifically, i don't know how to turn it into an impulse response that I can use live on the mic A to get mic B.

Could someone help be understand what this "equalizer" is and how i can turn it into an impulse response? Or maybe someone has another method to do this?

(btw, i tried to post on these questions, but as a new user, i can't comment yet... and i did not want to add an answer, that is not really an answer...).

  • $\begingroup$ What is the purpose of your ref microphone? And what are you actually trying to measure? I understand you think an intermediate step for whatever you're trying to measure is to get the impulse response from your ref mic to the others, but what is the actual final result you're looking for? $\endgroup$
    – Jdip
    Dec 1, 2023 at 20:11
  • $\begingroup$ I have a few different applications for this, number one being to simulate the stereo pair out of just one mic and an IR. Another would be to have just one mic pair and 9 stereo IRs instead of 10 mic pairs (thus dividing the project size by as many IRs i have. $\endgroup$
    – arno sica
    Dec 1, 2023 at 23:19

1 Answer 1


Nevermind, i figured it out, "equalizer" already is an IR... This might not be suited for what i want to do though because the calculation for that IR is proportional to the size of h... and matrix inversions for that size are pretty heavy, and i need longer IRs to be able to correctly reproduce the low frequencies changes.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.